Python_机器学习_李宏毅老师作业1

# By Richard
import sys
import pandas as pd
import numpy as np
import math

"""
sys：该模块提供对解释器使用或维护的一些变量的访问，以及与解释器强烈交互的函数
pandas：一个强大的分析结构化数据的工具集
numpy： Python的一个扩展程序库，支持大量的维度数组与矩阵运算
math：数学运算的库
作业一：由每天前9个小时的18个空气的影响因素（如：NO，CO，SO2，PM2.5等等）来预测第10个小时的PM2.5，train.csv是一年的数据，每个月取了20天，每天24小时
"""
data = pd.read_csv(r"G:\课程学习\机器学习\Mr_Li_ML\HomeWorks\数据\hw1\train.csv", encoding="big5")  # big5为了兼容台湾的编码集
data_train = data.iloc[:, 3:]  # iloc：通过行/列号选取数据 数据从第二行开始选择
data_train[data_train == "NR"] = 0
# print(data_train)
data_train_row = data_train.to_numpy()  # DataFrame 转换成numpy
# print(data_train_row)   #4320×24
# print(type(data_train_row))
# 将4320*24调整为12*18*480
month_data = {}
for month in range(12):
    sample = np.zeros([18, 480])
    for day in range(20):
        sample[:, 24 * day:(day + 1) * 24] = data_train_row[18 * (20 * month + day): 18 * (20 * month + day + 1), :]
    month_data[month] = sample
# print(month_data)
# 构建以将20天的480个小时看成连续的，所以一个月就有480-9=471个data，一年有471×12=5652个data，同样有5652个Label（第10个小时的PM2.5），采用这种方法可以构建较多的data
x = np.empty([12 * 471, 18 * 9], dtype=float)
y = np.empty([12 * 471, 1], dtype=float)
for month in range(12):
    for day in range(20):
        for hour in range(24):
            if day == 19 and hour > 14:  # 取消最后一天的往后延迟
                continue
            x[471 * month + day * 24 + hour, :] = month_data[month][:, day * 24 + hour:day * 24 + hour + 9].reshape(1,
                                                                                                                    -1)  # 转换成一行
            y[471 * month + day * 24 + hour, 0] = month_data[month][9, day * 24 + hour + 9]
# print(x)
# print(y)

# 归一化
x_mean = np.mean(x, axis=0)
x_std = np.std(x, axis=0)
for i in range(x.shape[0]):
    for j in range(x.shape[1]):
        if x_std[j] != 0:
            x[i][j] = (x[i][j] - x_mean[j]) / x_std[j]
# print(x)
# 将训练数据分成训练集-验证集，
x_train_set = x[:math.floor(len(x) * 0.8), :]    #4521
y_train_set = y[:math.floor(len(x) * 0.8), :]
x_validation = x[math.floor(len(x) * 0.8):, :]
y_validation = y[math.floor(len(x) * 0.8):, :]

# Training is readying
# 设置维度，存在偏差 bias  ax+b
dim = 18 * 9 + 1
w = np.zeros([dim, 1])
# 设置x_train_set 维度为4521*163
# 连接两个矩阵  列连接
x_train_set = np.concatenate((np.ones([len(x_train_set), 1]), x_train_set), axis=1).astype(float)
# set learing rate
learning_rate = 1
# ser iter time
iter_time = 30000
# set RMSprop parameters
adagrad = np.zeros([dim, 1])
eps = 0.00000001
# beta = 0.9
# iter is runing
for t in range(iter_time):
    loss = np.sqrt(np.sum(np.power(np.dot(x_train_set,w)-y_train_set,2))/len(x_train_set))
    # dot函数为numpy库下的一个函数，主要用于矩阵的乘法运算，其中包括：向量内积、多维矩阵乘法和矩阵与向量的乘法。
    if (t % 100 == 0):  #显示第一百次的结果
        print("迭代次数: %i, 损失值： %i" % (t, loss))
        # calculate gradient        transpose换了轴
        gradient = (np.dot(x_train_set.transpose(), np.dot(x_train_set, w) - y_train_set)) / (
                    loss * len(x_train_set))
        adagrad += (gradient ** 2)
        # reset parameters
        w = w - learning_rate * gradient / np.sqrt(adagrad + eps)

# save parameters
np.save("weight.npy",w)


# 验证
x_validation = np.concatenate((np.ones([len(x_validation),1]),x_validation),axis=1).astype(float)
for m in range(len(x_validation)):
    Loss = np.sqrt(np.sum(np.power(np.dot(x_validation,w)-y_validation,2))/len(x_validation))
print("The Loss on val data is %f"%(Loss))

# Author:Richard
import sys
import pandas as pd
import numpy as np
import math

testdata = pd.read_csv(r"G:\课程学习\机器学习\Mr_Li_ML\HomeWorks\数据\hw1\test.csv", header=None, encoding="big5")
test_data = testdata.iloc[:, 2:]
test_data[test_data == "NR"] = 0
test_data = test_data.to_numpy()
test_x = np.empty([240, 18 * 9], dtype=float)
mean_x = np.mean(test_x, axis=0)
std_x = np.std(test_x, axis=0)
for i in range(240):
    test_x[i, :] = test_data[18 * i:18 * (i + 1), :].reshape(1, -1)
for i in range(len(test_x)):
    for j in range(len(test_x[0])):
        if std_x[j] != 0:
            test_x[i][j] = (test_x[i][j] - mean_x[j]) / std_x[j]
test_x = np.concatenate((np.ones([240, 1]), test_x), axis=1).astype(float)
#print(test_x)
# Prediction
w = np.load("weight.npy")
#print(w)
ans_y = np.dot(test_x, w)
print("----预测PM2.5的值----")
print(ans_y)